Our study was one of the larger studies targeted at children under 1 year old in China. It demonstrated a high positive culture rate in infantile CNLDO. Among all the colonization, S. pneumoniae was the major isolate. Haemophilus and Neisseriae species were also commonly detected colonies. Chloramphenicol and levofloxacin were active agents for most of the pathogens in CNLDO, while erythromycin and sulfamethoxazole were proven to be resisted in a relatively high proportion.
There have been a few studies investigating the microbiology in CNLDO, and the bacterial spectrum varies among different age groups and changes over time (Table 3) [8-11, 13-18]. Most of the previous literatures included both infants and young children. Only 2 studies focused on infants under 1 year old, which were published 2 to 3 decades ago. The latter revealed a prevalent growth of Staphylococcus aureus (S. aureus) with a positive culture rate of 8.9-25%, and few evidence of S. pneumoniae infection (0-2%) [14, 16]. It was inconsistent with our observation, where S. pneumoniae constituted for a major proportion of 32.1% and MRSA only 7.1%. Though the details of topical antibiotics which the infants used before probing were not mentioned in these 2 literatures, we speculate that this could cause the differences of microorganisms detected in CNLDO .
Microbiota varies in different microhabitats of human eyes. The ocular surface, conjunctiva, lid margin and skin might show respective distinct bacterial spectra . The sampling location is vital to the analysis of microbiome in CNLDO patients. In most of the previous studies, the samples were obtained from conjunctival discharge by compressing LS. The specimen could be contaminated by the conjunctiva or lid margin, or sometimes little discharge with insufficient bacterial load could be obtained. In our study, we first sterilized the conjunctival sac, palpebral margin and skin, then collected the refluxed secretion from the lacrimal puncta by irrigation. This procedure would ensure maximum amount of specimen from LS.
Up till now, there have been 3 studies collecting irrigation samples of CNLDO, and they were investigated in Germany and Korea more than 10 years ago [9, 14, 15]. These literatures showed a high growth of S. aureus (13-25%), a low growth of Neisseriae (0.8-2%) and a variable clustering of S. pneumoniae (2-31%). The above were in contrast to our study which showed that Streptococcus and Neisseriae species were the most common, whereas S. aureus was a rare isolate. The difference could be attributed to microbiol changes with time, race or locality. Prokosch’s studies were 10 years later than Kim’s, and the former showed prevalence of S. pneumoniae which was similar to ours [9, 14, 15]. The Korean children Kim reported were closer to our patients with respect to race (both Asians) and locality compared to those in Prokosch’s studies conducted in Germany, which showed a much fewer isolation of S. pneumoniae. Prior antibiotic use might influence the result, and lead to a relatively low positive culture rate (64% vs 87.5% in our study, 97% in Prokosch’s studies) . Non-gonococcal, non-meningococcal Neisseriae were rarely reported in CNLDO. Antibiotic sensitivity tests of Neisseriae were not routinely performed in our hospital. Whether the bacteria are not susceptible to levofloxacin or tobramycin, or Chinese are genetically more susceptible to Neisseriae infection needs further study to prove.
S. aureus was one of the most common bacterial pathogens in neonatal conjunctivitis with a positive rate of 17-37.4% about 30 years ago [23, 24]. However, a study from southern China revealed a declining trend of S. aureus from 2002 to 2016, which was assumed to be attributed to antibiotics abuse . In our study, we advised patients with conjunctivitis to use levofloxacin every time they passed purulent discharge before probing could be performed, which might lead to no detection of S. aureus, and the result was consistent with that of A. Kuchar . So far, MRSA has been rarely isolated from infants with CNLDO. Sylvia Kodsi reported a case of MRSA cultured from the regurgitated pus in an 8.5-month-old child , while the other 2 cases demonstrated its overgrowth in conjunctiva and blood, respectively [27, 28]. MRSA could be related to chronic systemic antibiotics administration , recent hospitalization  or vertical transmission from the mother . Our study identified 2 infants with MRSA colonization in a total of 32 patients. No prior systemic antibiotics were used, no special signs of infection were detected in the infants and their family members, and their ocular symptoms resolved completely after an uneventful probing without causing any other infectious diseases.
H. influenzae and S. pneumoniae were reported to be prevalent bacteria in CNLDO patients with a wider range of age (Table 3), both pathogens can induce bacteremia after lacrimal probing [29, 30], and S. pneumoniae can cause severe endophthalmitis following glaucoma or cataract surgeries [31, 32]. Neisseriae species are part of the normal flora in respiratory system. Non-gonococcal, non-meningococcal Neisseriae are usually not pathogenic, but they can still lead to severe infections such as sepsis and endocarditis on occasion . Since the high prevalences of Streptococcus, Haemophilus and Neisseriae were reported in our study, empirical use of antibiotics against these bacteria should be considered as the initial treatment if the infants with CNLDO develop sepsis. Furthermore, the need of investigating the bacteriology after lacrimal probing should be emphasized. By knowing the prevalent bacteria and susceptibility results in CNLDO, it which could help prepare the doctors in treating the potential severe infections especially in young children.
Since bacterial conjunctivitis occurs intermittently in CNLDO, topical antibiotics should be given when a purulent discharge is present . It’s crucial to make prudent choice of antibiotic eye drops for infants, because they might require long term use of eye drops until the infection is treated. Chloramphenicol and levofloxacin were reported to be the most active according to our susceptibility test, and are usually used as topical eye drops instead of systemic antibiotics. Though literatures of infants under 1 year old are uncommon, levofloxacin is still deemed as an effective and safe antibiotic for infants with bacterial conjunctivitis [35, 36]. Despite the activity of chloramphenicol against most bacteria in CNLDO, its ineffectiveness has been reported in pediatric conjunctivitis . Furthermore, it is contraindicated in children with G6PD deficiency, which is common in southern China , and some doctors have concern for its possible side effects of gray baby syndrome and aplastic anemia in newborn [39, 40]. It is noteworthy that tobramycin, a commonly prescribed medicine in pediatric clinic , may sometimes be ineffective in CNLDO according to literature review (Table 3). Erythromycin, another frequently prescribed ophthalmic prescription , is proven to be ineffective for CNLDO, which is in accordance with most of the previous studies regardless of the year of study. Above all, topical administration of levofloxacin would be a better choice for Chinese infantile CNLDO with purulent discharge.
Pollard reported that 2.9% of infants with CNLDO would develop acute dacryocystitis , among which, 22.7% were concurrent with bacteremia . In such situation, antibiotics are usually given intravenously, including penicillins, cephalosporins, clindamycin, and vancomycin , which are consistent with our study.
The limitations of the study lie in the following aspects: 1. the preceding usage of antibiotics could influence the bacteria profile; 2. our hospital is a tertiary referral institution of CNLDO in Zhejiang which is a well-developed province in southern China. The hospital-based study represent the local condition, but might not represent the other areas in China; 3. the bacteriology of asymptomatic CNLDO might not be included; 4. variation of the bacterial spectrum with age was not certain; 5. retrospective design of the study hindered absolute standardization of intervention and data collection, which in turn reduce the level of evidence. However, the strength of this study is a large sample size focusing only on infants under 1 year old, and the unified treatment strategy and a reasonable way of collecting LS specimen. For the further study, a prospective, multi-centered investigation involving more patients of broader age spectrum with both symptomatic and asymptomatic CNLDO should be conducted.